Indian engineers develop 'Light' which could make you stop using Google search
Tall claims are everywhere. But when I tried out the Lightapp, there was substance in the claim. Developed by Indian techies who’ve had experience working in the Bay Area, Lightapp recently came out of stealth at the Gennext Innovation Hub, an accelerator powered by Reliance Industries and Microsoft Ventures. Post launch, we got an email from Sanjeev Nair, a technologist and design guy with close to two decades of experience who had something interesting to show.
A part of the core team, Sanjeev is also one of the investors in Lightapp. And as the tag line suggests, the app is about ‘going beyond search’. Light is an answering engine built on NLP (Natural Language Processing), machine learning and man-machine hybrid technologies. What does that mean? Well, you ask Light a question and it gives you an answer as if you’re having a conversation with it. No links, no searching for answers, the ‘bot’ does it all for you. Yes, there have been efforts on similar lines (including Google), and globally corporates have dedicated teams to work on these technologies making impressive headway, but Light has something noteworthy to show.
I tried out the app and the results were pretty impressive. I tried out a variety of questions, all of which would require different kinds of processing and the responses were accurate. Right from factual information like ‘what is the temperature right now?’ to ‘how deep do roots grow?’ to ‘what is the meaning of life?’, the app had answers that were satisfactory.
How does it work?
Well, one can throw in a lot of buzz words here but at the core, it is the implementation of a lot of work that has gone in the field of human-computer interaction. “It is system that processes a lot of data (big data), makes associations (relational data), learns (artificial intelligence/machine learning) and makes conversation with humans,” says Sanjeev.
Animesh Samuel and Sanjeev Menon are the key people behind the Light system. Animesh who is the CEO is an engineer who has consulting and marketing experience in India and the Middle East. And Sanjeev Menon is the CTO who after his MS at the University of Florida, and after working in a telecommunications startup for a while, had got a $22Mn exit with NetYantra Inc., a VoIP product company he founded in 2001.
The trio (Sanjeev Nair, Animesh Samuel and Sanjeev Menon) got together after a good 15 years in 2011 after knowing each other during engineering days. Starting out work after a discussion over some fine spirits, the app has been in the works for more than three years now. A lot of intelligence that the system possesses is a careful assimilation of the work that went in during these years.
Apart from Machine Learning and NLP, the system also relies on 'Man-Machine hybrid technologies'- a process by which Light manages the data that the machine does not comprehend, by supporting it with human input. For eg- a psychology based query that may need to have a more evolved answer, or an opinion based answer that needs to be sent out based on a topical query. The NLP system learns from each Human input it receives, thus facilitating future queries on the same topic to be auto-answered by the system. "We have the option wherein users of the app can become experts also, wherein they can answer queries related to their domain of expertise. Users can sign up as experts on the app. Expert answers are checked and moderated by the system as well as by peer review," says Sanjeev.
The journey
With a background in the AI space, the vision was pretty clear with Light. Sanjeev outlined this in a blog post where he mentioned that an average human being in the current age spends 47,520 hours (2.2 yrs) searching for information online. This beats the amount of time spent on things like the time spent in the bathroom (13148.7 hours which is 1.5 years) or laughing (2760 hours which is 115 days)! Very interesting stats and definitely a great pitch to anyone about why Light is needed.
They started work on the app and the initial idea was to have it underneath SMS. A user would send a message to the system which will throw back an answer. They built this and spread the word across their networks and organically they grew to a point where the system was processing about 100k queries a day.
Growing in confidence, and looking at the Internet and smartphone penetration, they conceptualized Light as an app and work began. Developing and designing it took a year and as of now, this Mumbai-based company employs 15 people. Boostrapped till now, the accelerator experience has opened up new doors for Light and they might soon raise a round of funding.
Moving ahead
With their launch on March 13, it is still very early days but the back story of Light is a powerful one. “Our sole focus now will be to take this to as many users as possible,” says Sanjeev. There are a lot of possibilities that can open up once they have a few million users considering the kind of data they’d be generating but those discussions are still far out to draw concrete conclusions.
I’ve had the app on the phone for a few days now and am returning regularly for asking questions. The traditional search is still required because a lot of our searches are for audio or documents or e-commerce but for many use cases, Light serves the purpose.
Try it out: Light App
